Project Description
The PI is working on cellular materials. Traditional means to measure material deformation is limited and the obtained information from traditional sensors is not sufficient. The PI has a 3D laser scanner with a fine resolution down to 0.2 mm. The laser scanner can record the material geometry in 3D with millions of data points. With the large data, the existing software can reproduce the material surface using a large number of triangles. We have been using visual examination to determine deformation, which is not efficient/accurate. We plan to develop computer codes to process the data such that the material deformation can be automatically obtained. We have three objectives: 1) identify the difference between a design object with a scanned object. This is similar to facial recognization, popular on modern cell phones except that we are looking for the difference rather than the similarity. 2) identify the difference between a deformed object with a undeformed object. This is possible with success in the first task. 3) determine material deformation using a series of scanned data during a physical test. This will likely reveal the research needs such as 3D markers. We will develop research proposals based on this work.
Tasks and Responsibilites
1) review of existing techniques/algorithms on comparing 3D point cloud data that describes an object. This is similar to the popular application of facial recognization and land surveys. 2) apply the existing techniques to the data obtained in the PI's research. This requires machine learning techniques such that the comparison can be feature-based rather than point-based. This is a key step. 3) develop algorithms to summarize features of a scanned object such that human interference can be minimized. 4) apply the algorithms to two set of scanned data to see if difference can be clearly identified. First we will focus on the perimeter of the object to determine the gobble deformation. 5) explore the use of markers placed on the object to improve the data process resolution and to obtain local deformation.
Desired Qualifications
None.